triattention

Here are 3 public repositories matching this topic...

atomicmilkshake / llama-cpp-turboquant

llama.cpp fork with TurboQuant quantization (turbo2/3/4) and TriAttention GPU-accelerated KV cache pruning. 75 tok/s on Qwen3-8B / RTX 3080.

windows cuda inference quantization kv-cache llm llama-cpp ggml turboquant triattention

Updated Jul 2, 2026
C++

atomicmilkshake / godzilla-llama.cpp

Star

MSVC+CUDA llama.cpp fork for Windows: KDA/GDN, long-context KV placement, TurboQuant/TCQ when measured, speculative draft. Lab defaults — TriAttention opt-in only, not recommended.

cuda kda kv-cache llama-cpp llm-inference speculative-decoding kimi-linear vram-optimization cuda-13 gated-deltanet turboquant triattention flashkda

Updated Jul 28, 2026
C++

Unitflexmed1821 / FlashKDA

Star

Accelerate Kimi Delta Attention computations with high-performance CUTLASS kernels designed for NVIDIA SM90 architectures and beyond.

cuda kimi kda moonshot kv-cache llama-cpp llm-inference speculative-decoding moonshot-ai kimi-k2 kimi-linear vram-optimization cuda-13 gated-deltanet turboquant triattention kimi-k3 flashkda

Updated Sep 3, 2026

Add this topic to your repo

To associate your repository with the triattention topic, visit your repo's landing page and select "manage topics."

Learn more